UPTEC Q 18022
Examensarbete 30 hp Juni 2018
Ageing tests of cemented carbide powders
An investigation for increased quality of metal cutting inserts
Eric Rösth
Teknisk- naturvetenskaplig fakultet UTH-enheten
Besöksadress:
Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0
Postadress:
Box 536 751 21 Uppsala
Telefon:
018 – 471 30 03
Telefax:
018 – 471 30 00
Hemsida:
http://www.teknat.uu.se/student
Abstract
Ageing tests of cemented carbide powders
Eric Rösth
In this study, the ageing effects on powder used for cemented carbide insert production are examined. Ageing is throughout this study, defined as the time dependent change of the magnetic properties: coercive field strength and saturation magnetization. Testing is done using eight different powder compositions stored in both air and in an argon cabinet for 10 weeks, where sampling is done at specific intervals. Samples are stored in vacuum sealed bags for a combined sintering at the last phase of the test. Magnetic properties are assumed to be dependent on the amount of oxides needed to be reduced by taking carbon from the material itself during the vacuum stage of the sintering.
To achieve interpretive results, this study also tested available sintering furnaces (DMK and DEK) by sintering trays with patterns of test pieces. This shows that DEK furnaces are much better for the ageing tests performed in this study, since less variation of the magnetic properties are measured because of the symmetrical heat gradient over each tray.
Ageing tests strongly suggest that the cause of ageing comes from water absorbed by the PEG in the powder composition. Changing the molecular weight of the PEG seems to have an effect on the powder's ageing sensitivity. Measurements performed in this study show less ageing for Cr-rich DA-powders than for cubic carbide rich DQ-powders.
Handledare: Bodil Lindberg och Sead Sabic, SANDVIK Coromant
Åldringstester av hårdmetallpulver
Eric Rösth
Effektivisering är ett ord man ofta hör inom all sorts industri. För lönsamhetens skull så måste material och råvaror användas smartare och det måste slösas mindre. Hårdmetallindustrin är inget undantag och därför har ett projekt genomförts i samarbete med SANDVIK Coromant där fokus ligger på hur lagringstidsberoende kvaliteten på det tillverkade hårdmetallpulvret faktiskt är.
Hårdmetaller används för tillverkning av skär för metallbearbetning. Detta sker genom att en mald pulverblandning av volframkarbider, kobolt, pressmedel och andra tillsatsämnen pressas till önskvärd form för att sedan fogas samman via sintring i stora ugnar vid höga temperaturer.
Hårdmetallerna som tillverkas på SANDVIK Coromant i Gimo är av världsklass. Därför är också kraven på fortsatt kvalitet och utveckling i produktionen mycket höga. Kassering av pulver som inte håller måttet sker ibland när dess egenskaper är utanför toleransen för den förväntade kvaliteten. Åldrande av pulver kan göra att pulvret hamnar utanför toleransen, baserat på både hur det lagrats och hur länge det lagrats.
Det här projektet omfattar en serie försök där slutmålet är att kartlägga förekommande åldringseffekter hos pulvret. Mycket tyder på att det är absorberat vatten i pressmedlet som orsakar uppmätta skillnader på kvaliteten för sintrat material. Skillnader har mätts över en tioveckorsperiod. Provbitar som lagrats i argon under samma tid uppvisar mycket lägre åldringseffekter. Detta då argonskåpet är mycket torrare än luften utanför.
För att styra egenskaperna på de blivande hårdmetallskären så mäts de magnetiska
egenskaperna efter sintring. Materialets magnetiska egenskaper kan användas för att utvärdera storleken på kristallkornen som materialet består av samt dess kolhalt. Dessa egenskaper är starkt kopplade till hur väl hårdmetallen fungerar för metallbearbetning. Detta blir då ett mått på vilket skick pulvret är i och hur väl utförd sintring som gjorts.
Beroende på vilka tillsatsämnen eller vilken sintringsprocess som använts så upplevs åldringseffekter olika tydligt. Den största åldringskänsligheten kommer från vilken sammansättning av pressmedel som använts.
Det som händer är att kolhalten går ner över tid. Detta troligtvis på grund av att provbitar absorberat mer vatten desto längre de varit exponerade mot omvärlden. Vatten i proverna förångas i början av sintringsprocessen vilket oxiderar ytorna i materialet. Dessa oxider avdrivs senare i sintringen genom att bindas med kol från materialet självt. Dessa avdrivs då som kolmonoxid.
För att säkerställa mätbara resultat har tillgängliga sintringsugnar noggrant kartlagts utefter deras prestation. Detta genom att utveckla en ny metod, lämplig för att utvärdera ugnarna.
Denna metod utgår från att fylla ugnar med flera provbitar av ett välkänt material. Efter sintring mäts bitarna, vilket då ger svar på hur mycket det skiljer sig inom ugnen men också mellan ugnarna. För ålderstesternas skull så behövdes en ugn som kunde leverera ett jämt
värde på många provbitar, eftersom ålderseffekten annars var lägre än variationen inom ugnen. Denna metod skulle kunna användas som en standard för att bestämma ugnsprestanda.
I framtiden kan det komma att bli nödvändigt att se över vilket pressmedel och vilka tillsatsämnen som används, men också vidta rejälare åtgärder mot luftfuktigheten i
produktionsanläggningarna. De produktionssteg som då medvetet innehåller vatten, så som malning och spray-torkning blir också extra viktiga att optimera om åldringseffekterna vill minimeras.
Examensarbete 30 hp på civilingenjörsprogrammet
Contents
1 Introduction 1
1.1 Background . . . 1
1.1.1 Definition of ageing . . . 1
1.2 Purpose . . . 2
1.2.1 Mapping of practical consequences . . . 2
1.2.2 Underlying mechanisms . . . 2
2 Theory 3 2.1 Magnetic properties and ageing . . . 3
2.2 Chemistry of ageing . . . 6
3 Method development 7 3.1 The furnace problem . . . 7
3.2 Position testing of DMK furnaces . . . 7
3.3 Position testing of DEK furnaces . . . 8
3.4 Results of position tests . . . 9
3.5 Discussion of position tests . . . 11
4 Method 12 4.1 Ageing tests . . . 12
4.1.1 Part 1 (late stage ageing) . . . 12
4.1.2 Part 2 (early stage ageing) . . . 14
4.2 Humidity testing . . . 15
5 Results 15 5.1 Part 1 (late stage ageing) . . . 16
5.2 Results of humidity tests . . . 19
5.3 Part 2 (early stage ageing) . . . 21
6 Discussion 21 6.1 Humidity . . . 21
6.2 Discussion of ageing results . . . 21
6.2.1 Part 1 (late stage ageing) . . . 21
6.2.2 Part 2 (early stage ageing) . . . 23
7 Conclusions 23
8 Further research 24
1 Introduction
1.1 Background
When producing powder that later will be used to produce cemented carbide for metal cutting, it is of great interest to gain control over WC grain size and carbon content, since controlling these are crucial for the quality of the final product. When a new pow- der is milled these parameters are indirectly verified by testing the magnetic properties of the pulp. If acceptable, the powder is dried, tested again and then packaged.
Cemented carbide is a composite material made out of tungsten carbides cemented together with cobalt as a binder. The material also contains supplementary elements to optimize its properties. The magnetic properties of cobalt are used to control and judge the achieved material properties of the cemented carbide. The powder also con- tains polyethylene glycol (PEG) to assist during pressing of the green bodies. PEG is degassed during the debinding step of the sintering.
1.1.1 Definition of ageing
The term ageing is often used with a wide spectrum of loose definitions. Therefore ageing in this study is defined as the change of the magnetic properties (Hc and CoM) over time, no matter the reason. Hc (kA/m) is the coercive field strength, which is the strength of an applied field needed to demagnetize the material. CoM (wt%) is a measurement of the magnetic saturation of the Co relative to pure Co, measured by weight. Not all cobalt in the material acts magnetic.
As soon as the powder is exposed to air it will start to age and the main cause of change is oxidation, especially together with heat [1]. It is often assumed that this is mainly because of oxygen in the air, but could just as well be because of humidity, or at least partially. Humidity is poorly controlled at some areas at the Gimo site where this study is performed.
Earlier reports suggest that long duration ageing causes Hc to increase and CoM to decrease [2][3]. These studies also suggest that ageing occurs very slowly in a controlled environment. Since not a lot of data exists on early stage ageing, it is possible to sus- pect that there are other mechanisms behind ageing of newly produced powders.
1.2 Purpose
1.2.1 Mapping of practical consequences
The aim of this study is to investigate the rate of ageing and how it differs between different powder compositions. For example, it is known that adding more Co binder and cubic carbides, like TiC and (Ta, Nb)/C, reduce the rate of late stage oxidation, while substituting Co for Ni makes oxidation resistance worse [4].
1.2.2 Underlying mechanisms
This study will also examine the underlying mechanisms of the changes in Hc/CoM and how well these changes actually translate into the more desirable properties of WC grain size and carbon content.
A prime example of the effects of ageing was witnessed during normal production of a standard powder at SANDVIK, where a set of pre-dried pulp tests were delayed for four days openly in air before being sintered. Since the test were delayed for four days, they were double checked using pieces from the same batch stored in an argon cabinet.
Differences between the pieces can be seen in table 1, using the same parameters that this study will be using to judge ageing [5].
Table 1: Effects of delayed sintering of a pulp test (mean values for a set of four test pieces each)
4 days in argon cabinet 4 days in air
Hc (kA/m) 22.67 22.73
Stdev Hc 0.021 0.021
CoM (wt%) 5.78 5.69
Stdev CoM 0.007 0.014
How is it possible to improve production with better knowledge on how the powders behave over time and at what point during production do ageing matter the most? To answer these questions, this study is split in two parts to investigate early and late stage ageing, respectively. In the process it will be necessary to map the furnaces in order to correctly judge changes in Hc/CoM.
2 Theory
2.1 Magnetic properties and ageing
As mentioned earlier, ageing will be judged upon two parameters, Hc (kA/m) and CoM (wt%) both of which are present in table 1. These parameters are magnetic properties and are somewhat connected to each other, which is visualized in figure 1.
Figure 1: Hc/CoM-relationship for Aa-powder batches from database (2017-01-01 to 2018-02-01).
The effect that time has on the cemented carbide is that there is a variance in the amount of stable oxides needed to be reduced when the green body is finally sintered.
Oxides are reduced during the vacuum stage of sintering using carbon from the material itself, which has an impact on carbon content. This in turn determines the amount of tungsten in the cobalt phase which lowers the magnetic saturation, causing it to become non-magnetic and therefore lowering CoM.
The lower carbon content and thus higher tungsten content increases the melting tem- perature and thereby reduces the WC grain size [6]. This causes pinning of magnetic domain walls which can be seen in figure 2. This is why Hc is connected to the grain size, since grain boundaries acts as an obstacle to these domain walls. Pinning is ba-
Figure 2: Pinning energy as a function of a cylindrical defect with a diameter D.
Showing both analytical values and simulated (FEMME). The pinning energy is the difference in energy between state A and B to the right. Being pinned means the defect is part of the domain wall. [7]
There are however other ways of pinning magnetic domain walls and there are also other ways of rendering cobalt non-magnetic, without changing carbon content. The magnetic properties of cobalt is for example highly dependent on the direction of crys- tallization. This dependence is shown in figure 3.
Figure 3: Direction dependency of magnetization for cobalt. [8]
Cubic carbides in the material are believed to protect against ageing. This is because the cubic carbides behave substoichiometric, which creates a γ-phase of which acts as a carbon buffer [9]. Which gives a less carbon sensitive powder.
When carbon leaves through degassing and reduction of stable oxides during sintering, it is possible for tungsten atoms to enter this cubic carbide structure. This means that tungsten will not be dissolved in the cobalt phase to the same extent and as mentioned earlier, the mechanism of ageing is tungsten in the cobalt phase after sintering. Carbon content would however be lowered, without a change in Hc/CoM.
2.2 Chemistry of ageing
The chemistry of sintering is discussed in depth in Jenny Angseryd’s ’Debinding of cemented carbide’ [10]. The most important points for the ageing tests performed in this current report are summarized below.
Early during sintering at relatively low temperatures, cobalt oxides are reduced using hydrogen gas from the surrounding atmosphere. This follows the reaction:
CoO+ H2 →Co+ H2O
This means that depending on the amount of cobalt oxides there will be an increase in the humidity of the hydrogen gas in the furnace. At the same time polyethylene glycol (PEG) in the green body can contain a lot of water which further can increase the humidity when degassed [11].
Depending on the humidity of the hydrogen gas, two other reactions take place when reducing more stable oxides (mainly oxides from tungsten but also Cr, Ti, Ta, Nb and other similar additions). These reactions are:
A: W O + 2C → W C + CO B : W O + H2+ C → W C + H2O
Because of the equilibrium with the atmosphere reaction A appears to be more dominant in a humid environment and can also happen freely in the vacuum stage of sintering.
The opposite is true for reaction B. Both reactions will take carbon from the material itself (A more than B) which will be lowering CoM after sintering.
This is theorized to be the underlying mechanics for ageing of cemented carbide. This is dependent on the amount of cobalt oxides, stable oxides and water in PEG. All of which could in turn be dependent more or less on time or atmosphere. A piece filled with a lot of water could potentially age other pieces during sintering if there is a steam cloud around the piece during the hydrogen step of sintering.
3 Method development
3.1 The furnace problem
A major concern for Hc/CoM measurements on a large amount of test pieces is the dispersion based on position in the sintering furnaces. Variances are caused by temper- ature gradients and differences in gas flow. In this study it is believed that this noise to the measurements have the possibility to completely drown any measured ageing effects.
To combat this, knowledge of available sintering furnaces (DMK and DEK) at the Gimo production site was needed. Therefore a way of testing the performance of the furnaces was developed, specifically for this study. DMK furnaces are the standard furnaces used at the Gimo production site.
If none of the furnaces in the production showed acceptable performance, a backup plan was to use the DDK lab furnace. This however, would mean a need to scale down the planned ageing tests since the DDK furnace is much smaller than the furnaces available in production.
3.2 Position testing of DMK furnaces
Because of the known dispersion caused by the position and unevenness of the DMK furnaces, a set of tests was performed to document what noise the furnaces could pos- sibly contribute to Hc/CoM-measurements.
For this test, the DMK1 furnace was used and a tray containing sintering control pieces (Sc) together with leftover Aa2/Aa3 pieces, was placed on top of the stack in the corner closest to the door of the DMK furnace with test pieces in the pattern visualized in figure 4. This was then sintered using the DA sintering process.
The DMK furnace contains six stacks of trays, all loaded from the front and the radi- ant section around the stacks is a cuboid. The specific tray position in the furnace was evaluated because it was the position used to sinter the early stage ageing tests that can be found in A26-A33.
An identical test was repeated for the DQ sintering process but with only Sc pieces.
Figure 4: Placement of the tray in the furnace seen from above. Trays were placed in the top of the stack with the first and last sets of samples oriented towards the middle of the furnace.
3.3 Position testing of DEK furnaces
Similar DEK tests were performed for both the EA and EQ process (DEK equivalent to DA/DQ), where the same pattern of Sc pieces on three trays were used. These three trays were placed inside the total DEK stack on the bottom, in the middle and on the top. The DEK1 furnace was used for these tests.
The DEK furnace’s radiant section is of a cylindrical shape, where a single stack of trays is elevated up into the furnace. The reasoning behind the usage of three plates was to also evaluate the differences based on height.
Later during the sintering of the part 1 late stage ageing tests, additional DEK height tests were performed were sintering control pieces were placed on every tray in the lower half of the stack.
3.4 Results of position tests
A large angular position dependency of Hc in DMK furnaces was found. However, DEK furnaces show no such dependency. No angular dependency of CoM in neither DMK nor DEK furnaces. This can be seen in figure 5-7.
The DEK test shown in figure 6 and 7, together with the additional test of height differences shown in figure 8, demonstrate that DEK furnaces have an large exponen- tial increase of Hc and a slight decrease of CoM with increasing height inside the furnace.
Based on available standard deviations a radial dependency of Hc and CoM exists on all trays. Variations between pieces on a single tray appear higher towards the bottom of the DEK furnace. More position data can be found in appendices A1-A7.
Figure 5: Position tests with Sc pieces in DMK1 DA. Orientation like in figure 4.
Figure 6: Hc results of position tests with Sc pieces in DEK1 EA. Three trays with test pieces on different levels of the total stack. Error bars show one standard deviation.
Figure 7: CoM results of position tests with Sc pieces in DEK1 EA. Three trays with test pieces on different levels of the total stack. Error bars show one standard deviation.
Figure 8: Hc/CoM for samples on 16 trays stacked vertically from the bottom of DEK1 EA. From the same sintering as part 1 EA test pieces.
3.5 Discussion of position tests
From position testing of the DMK furnaces it is impossible to achieve a fair pattern on the trays at the specific location tested. There is at least a 0.2 kA/m difference in Hc based on the angular position in the DMK furnace. CoM seems to have no dependency on angular position. The main reason for the Hc differences is the asymmetrical heat gradient, caused by the geometry of the cuboid radiant section of the furnace together with the six stacks of circular trays. Because of this, it is reasonable to believe that any other position of the test tray would still give large variations in Hc.
The angular dependency presented in figure 5 can be seen in the Hc measurements of the early stage ageing tests presented in appendices A30-A33. Because of this, no valuable ageing data could be seen, but it proves that the DMK furnace is the improper choice of furnace for ageing tests. If DMK furnaces are used for testing, ageing can only be judged upon CoM.
In comparison, the DEK furnace performs much better for this type of testing since all pieces on one radius become comparable with their respective piece on another angle.
This is because the symmetry of the DEK furnace. Standard deviations show that there is a radial gradient over the trays in DEK.
The DEK furnace shows a large difference in performance based on height. This means trays in the DEK stack, at least become comparable with the tray itself and those trays in its nearest vicinity. The height variations are believed to exist because of the differences in gas flow and purity of gas throughout the DEK furnace.
The DEK furnaces appear to give comparable results (according to position on tray) to the lab DDK furnace from prior studies [2]. DDK furnace should have a much smaller variation based on height and also a lower radial dependency.
4 Method
4.1 Ageing tests
4.1.1 Part 1 (late stage ageing)
The first ageing test of this study was initiated by opening eight containers of different cemented carbide powder compositions from the inventory. These different powders were then pressed into eight batches, containing 288 pieces per batch. Pressing was done utilizing the same press and tools. Pieces pressed were CNMG 12.
Table 2: SANDVIK-coded table of the specific powder containers opened for the test
DA-powders Aa Ab Ac Ad
DQ-powders Ba Bb Bc Bd
It should be noted that all DQ-powders contain a large amount of cubic carbides com- pared to the DA-powders. In addition, the Bb powder also contain a different composi- tion of PEG compared to all the other powders used in this test. DA-powders contain a few percent chromium.
Half of each group of 288 pieces were placed inside the argon cabinet and the other half was placed openly in air. Sample collecting was done by picking eight pieces from each batch, four from the pieces in the air and four from those in the argon cabinet. A set of four pieces represented one point of measurement which in this study is referred to as one sample.
Figure 9: One of the eight trays stored in air.
Samples were then vacuum sealed very carefully to ensure ageing would slow down drastically. They were sealed together based on the DA/DQ-split from table 2. Sealed bags of samples were then stored waiting to be sintered. The bags used for sealing were metallurized polymer bags. Bags are seen in figure 10.
Figure 10: Storage of metallurized polymer bags containing samples.
All sintering in part 1 was done using the DEK1 furnace, which has one stack of trays elevated up into the furnace itself. Sintering was done after a 10 week period. For approximately one hour before sintering, all bags were opened and pieces were put on trays in specific patterns. Picture of trays can be seen in figure 11. The test stack of trays were put in the lower half of the DEK furnace. EA and EQ sintering procedures were used (DA/DQ equivalent for DEK) based on the table 2 split. One sintering con-
Figure 11: Pattern of pieces on tray post-sintering. The sets of four based on angle on tray represents one sample.
The chosen intervals of measurements are visible in the results below. The main idea was to start with very short intervals and then move on to longer ones as the test pro- gressed. Measurements were done after sintering using an automated Hc/CoM-robot, Koerzimat Robotics from Foerster. All measurements in this study were done using this machine.
4.1.2 Part 2 (early stage ageing)
The second part started when the batch Aa2 was dried using normal means of produc- tion. A few kg of powder was taken directly after the drier were done. Aa2 here is same powder as Aa from table 2 but the number indicates another batch.
Some of the powder was directly pressed into pieces. The remaining powder was split into five containers. One container with air flow that was shaken between sampling, one closed container standing in air and three closed containers standing in the argon cabinet. Pressing was done using the SNUN lab press.
Sampling was done by taking four pieces from the pre-pressed ones and pressing four additional pieces for each of the mentioned containers in air. Each container from the argon cabinet were brought out for extra samples at three different occasions. All sam- ple pieces were then vacuum sealed in metallurized polymer bags to stop ageing the same way as were done with the samples in part 1.
The intervals of sampling were much shorter than in part 1 and sampling stopped after 45-50 hours. Bags were then opened and pieces were placed on trays for sintering. Pat- tern on the trays were the same as in figure 11 but DMK furnaces were used instead.
Sintering in DMK was always done with the trays placed in the top of the stack and near a corner of the furnace. DMK furnaces are horizontally loaded with six stacks of trays.
This was repeated with the batches Aa3, Bd2 and Bd3 but with the addition that the containers in the argon cabinet were left out in the air after they had been brought outside. These containers were additionally tested with short intervals the first few hours they spent in air.
These three repeated tests also had the orientation of the trays controlled for (not only the pattern on the tray as was the case of the Aa2 test). Figure 4 shows how the trays for Aa3, Bd2 and Bd3 were rotated in the furnace, with the first and last sets of sample pieces next to each other. The same height and position were used in all four tests. Aa and Bd used different sintering procedures (DA/DQ) and tests were done at different times. Meaning different DMK furnaces were used. Measurements were done after sintering.
4.2 Humidity testing
10 weeks old Bb and Bd pieces were taken from both the air stored and the argon cabinet stored trays. Some of the pieces were placed inside an oven to be dried at 120
◦C for an hour. Some other pieces were treated with a few drops of water, which on a scale represented around a 0.2 gram weight increase for each test piece.
These pieces were then put on trays in specific patterns. The patterns isolated test pieces into the categories ’wet pieces’, ’wet next to untreated pieces’, ’untreated pieces’
’untreated next to dry pieces’ and ’dry pieces’. These trays were then sintered in an otherwise empty DEK furnace to further isolate test pieces from other contaminating sources.
5 Results
All results not presented here can be found in the appendices A9 to A33. In this results section, specific examples of interest will be given as a basis for later discussion.
Examples will be given along with standard deviations.
5.1 Part 1 (late stage ageing)
The summary of Hc/CoM results (together with standard deviations) for storage of pieces in air can be found in figure 12. Figure 12 is a combination of the first and last values measured from appendices: A8, A10, A12, A14, A16, A18, A20 and A22.
Samples stored for 10 weeks in air has an increase in Hc for seven out of eight of the powder compositions tested, where at least five powders has an increase significant in comparison to the standard deviation.
All samples stored in air had a decrease in CoM, although all CoM values are much closer to their respective standard deviations. DQ-powders show a larger sensitivity towards ageing than DA-powders.
Figure 12: Total difference in Hc/CoM acquired over 10 weeks storage in air. Graph shows the difference between the first and the last samples for each powder. Inserted graph shows one standard deviation for the four pieces in one sample.
The summary of Hc/CoM results (together with standard deviations) for storage of pieces in argon can be found in figure 13. Figure 13 is a combination of the first and last values measured in A9, A11, A13, A15, A17, A19, A21 and A23.
Samples stored in the argon cabinet for 10 weeks show small differences in both Hc/CoM.
No general trend between the different powder compositions in terms of increase/decrease of Hc/CoM. Bb pieces do show a large decrease in Hc when stored in argon. Larger standard deviations for argon stored samples were observed compared to the air stored samples.
Figure 13: Total difference in Hc/CoM acquired over 10 weeks storage in the argon cabinet. Graph shows the difference between the first and the last samples for each powder. Inserted graph shows one standard deviation for the four pieces in one sample.
Starting values between air and argon stored samples of the same powder composition showed some variances. These variances have been compiled in figure 14 for Hc and figure 15 for CoM.
Bb pieces showed a large variance in Hc between the initial samples stored in air and argon. Figure 14 shows that there is no general trend between the powders in term of positive/negative values.
Figure 14: Air/argon Hc differences for the first samples (4h). Inserted graph shows one standard deviation for the four pieces in one sample.
Ac pieces showed a large air/argon starting value variance in CoM. Figure 15 also shows that there is no general trend for CoM in terms of positive/negative values just as there was not a trend for Hc in figure 14. Differences between the graphs however, seem to follow that when CoM drops, Hc increases and vice versa.
Figure 15: Air/argon CoM differences for the first samples (4h). Inserted graph shows one standard deviation for the four pieces in one sample.
5.2 Results of humidity tests
Figure 16-17 shows that adding water causes Hc to increase and CoM to decrease.
Pieces with different moisture levels appear to not be influencing each other. Pre- drying samples before sintering caused no difference. Additional humidity results for Bd pieces can be found in A24-A25. Humidity test graphs also show the total Hc/CoM difference acquired over 10 weeks for the untreated samples.
Figure 16: Hc results for humidity tests on Bb pieces. Error bars show one standard deviation.
Figure 17: CoM results for humidity tests on Bb pieces. Error bars show one standard deviation.
5.3 Part 2 (early stage ageing)
Hc measurements in DMK gives a large amount of noise. CoM-curves appear to behave the same between the different repeated tests for each powder, respectively. This was true despite A26 and A27 having different orientations. Graphs can be found in A26- A33
6 Discussion
6.1 Humidity
The oven used in an attempt to dry the samples before putting them into the sintering furnace either didn’t achieve its goal or drying the samples didn’t have an effect. How- ever at the same time, actively putting water into the samples before sintering had a clear measurable impact on both Hc and CoM.
The fact that water had such an impact points towards that water in the PEG or water trapped in the green body through capillary action, is the cause of ageing effects. The attempt to dry the samples did not affect the results which indicates that drying of H2O dissolved in PEG must be studied further.
Difference between argon stored and air stored samples in the humidity tests found in figure 16 and 17 (untreated columns) also show the total difference in Hc/CoM acquired over the 10 week test period, independent on the vacuum sealing that was done in the part 1 late stage ageing tests. Only the samples that had water directly put in them showed a large difference caused by the water and did not affect other samples in their proximity. If there is a steam cloud around the samples during sintering at all, it is way too small and degassed way too quickly.
6.2 Discussion of ageing results
6.2.1 Part 1 (late stage ageing)
The trend of samples stored in air appears to be that Hc increases and that CoM de- creases. Which then would infer that the amount of cobalt oxides, stable oxides or water in PEG was higher in those samples before sintering. The chromium rich DA- powders seem to have a better protection towards ageing effects than the cubic carbide rich DQ-powders. It is unclear if measured differences are because of the additives or the sintering process. Cubic carbides also have the possibility to hide ageing effects (as defined in this report) because of the ability to solve tungsten atoms into the cubic carbide structure, meaning carbon can leave without lowering CoM [9]. Which is what would happen if oxides are reduced with carbon from the material but the interstitial
The rate of CoM change was lower than in prior study [3]. That study did however not vacuum seal its samples but instead sintered them directly after sample collection. This could mean that the humidity of the local weather is locked for each sample. Inter- estingly, the cited study started during winter/spring, which means that local weather has an increase of humidity as the testing proceeds. This could mean that ageing simply is an effect of the weather or seasonal variances in humidity, if it is caused by moisture in the samples. This method of sintering each sample directly after sampling would not be possibly at the Gimo site, since the furnaces differ too much between runs.
This thesis study also started during winter but did instead vacuum seal samples for a combined sintering at the end. Ageing effects were still measured. This could mean that local air humidity does not create an equilibrium with the samples instantaneously.
It is also possible that the samples never reached an equilibrium with the air humidity, but rather kept absorbing moisture from the air for as long as they were exposed to it.
Vacuum sealing samples could have an effect on absorbed moisture. It is possible to believe that the vacuum might have a drying effect on the samples simply by creating another equilibrium. Vacuum sealing is however not ideal and some air will enter the bag. Air that enters the bag might contribute to ageing which would then mean that this thesis study also just locked the daily weather humidity during sample collecting.
Variation of quality of vacuum seals could be the reason for the variation of the early samples.
The argon cabinet appears to stop ageing better than the vacuum sealed bags. Espe- cially for the more sensitive Bb-powder (different PEG composition). This is believed to be because the atmosphere in the argon cabinet is drier than the achieved vacuum.
This shows as Hc decreasing and CoM increasing, since the first samples has been stored in vacuum bags longer and therefore aged more. For some powders, there is a noticeable difference between air and argon storage for the first samples. This could mean that something happens very rapidly, since the first sampling in this test was done after four hours.
Some of the powders show very different ageing compared to earlier studies [2][3]. Know- ing this, together with the variances in performance of the argon cabinet and the direct proof of Hc/CoM change based on moisture in the samples, it is very unlikely that witnessed ageing effects are caused by anything else than water absorption of the PEG.
This is further strengthened by that the Bb-powder containing the largest fraction of the most hydrophilic PEG had the largest variation. Both the speed of water absorp- tion and the amount of water absorption should play a role in ageing effects.
6.2.2 Part 2 (early stage ageing)
The Hc graphs in appendices A30-A33 from the early stage ageing tests contain too much noise to see any time dependency. This was discovered from the position tests for the DMK furnaces. CoM graphs in appendices A26-A29 show that Aa powder stored in air flow has an increase in CoM. This could be another drying effect since the spray-drier used in production has some variation in performance. Tests performed to map early stage ageing do however have a major uncertainty since the Hc graphs are discarded from because of the already mentioned noise.
These short term variations could be interesting since the powder is tested before and after the spray drier to ensure the quality is correct. Variations of time between testing could be a few hours or even days which could result in acceptable powder being thrown away or vice versa.
7 Conclusions
Position tests show that the DEK furnace is the better choice for this type of testing, where a high number of samples need to be sintered and compared to each other based on Hc/CoM measurements. The method developed for these position tests works and could be made into a standard for furnace evaluation.
Humidity tests and ageing tests strongly suggest that water in the PEG or green body is the main cause of measured ageing effects. Variation in PEG composition seem to have an effect in sensitivity towards ageing. These tests also suggest that the argon cabinet is effective at slowing down time-dependent change and in some cases even reverse it.
No steam cloud during sintering that affected other pieces was found and drying pieces in an external oven did not work.
In this study, the Cr-rich DA-powders aged less than the cubic carbide rich DQ-powders.
It is unclear if this is because of the additives or the sintering process. DQ-powders might have lost some carbon content which do not show as a difference in CoM.
8 Further research
To further research and to verify tests performed in this study, more controlled test could be performed. Tests with control over humidity could be done to verify. To see if its the local weather, the tests could also be repeated during autumn. This would rule out all seasonal dependence.
To see if its possible be restore powders by drying them, tests should be done where green bodies are held in a dry hydrogen atmosphere at low temperature (100-120 ◦C), for a period of time directly in the sintering furnace, before the regular sintering process.
It should be evaluated if the measured ageing differences between DA/DQ-powders de- pends on the sintering process with this drying theory in mind.
The early stage ageing tests performed in this study should be repeated with the DEK furnace instead of the DMK furnace. This could give valuable information of how time sensitive quality testing of pulp and spray dried powder actually is. These tests could possibly also answer why there is a noticeable difference between air and argon for the first samples done in the late stage ageing tests of this study (figure 14 and 15).
Different PEG variations should be tested, with chemical analysis before and after sin- tering. The ageing sensitivity caused by PEG should be evaluated, where the speed and the amount of water absorbed should be considered. There is also a need to figure out if the PEG actually reaches an equilibrium with the air humidity or if there is an constant absorption of available water. Testing the weight of samples over time could give data on absorption speed.
Specific tests with chromium carbide and cubic carbide additives could be performed to see if these additives are relevant to ageing effects.
References
[1] A.S. Kurlov and A.I. Gusev. Oxidation of tungsten carbide powders in air. Int. J.
Refractory Metals and Hard Materials, 41:300–307, 2013.
[2] N. Karam. Aging of Powder Samples. Technical Report SANDVIK Coromant, 2017.
[3] M. Waldenstr¨om. ˚Aldringsf¨ors¨ok Cr-sorter. Technical Memo SANDVIK, 2011.
[4] C. Barbatti, J. Garcia, P. Brito, and A.R. Pyzalla. Influence of WC replacement by TiC and (Ta,Nb)C on the oxidation resistance of Co-based cemented carbides.
Int. Journal of Refractory Metals & Hard Materials, 27:768–776, 2009.
[5] Internal document SANDVIK Coromant. Sammanfattning pulpprovdokument.
[6] O. Bj¨areborn. WC grain size distribution during sintering of WC-Co cemented carbides. Lund University Publications, 2016.
[7] D. Punz, J. Lee, M. Fuger, and D. Suess. Theory and micromagnetics of pinning mechanism at cylindrical defects in perpendicular magnetic films. Int. Journal of Applied Physics, 107, 2010.
[8] B.D. Cullity and C.D. Graham. Introduction to magnetic materials, second edition (page 202), 2009.
[9] G. Andersson and B. Jansson. THE SOLUBILITY OF CUBIC CARBIDE FOR- MERS IN LIQUID COBALT. International Plansee Seminar, 2:662–676, 2001.
[10] J. Angseryd. Study of the chemical reactions during debinding of cemented carbide.
Laboratory Report SANDVIK Coromant, 2004.
[11] I. Osman, K. Seyfullah, and C. Burcu. The Effect of PEG on the Water Absorption Capacity and Rate of Superabsorbent Copolymers Based on Acrylic Acid. Int.
Journal of Polymeric Materials, 54:1001–1008, 2006.
Appendix
Figure A1: Position tests with Aa2 pieces in DMK1 DA. Orientation like in figure 4.
Figure A2: Position tests with Aa3 pieces in DMK1 DA. Orientation like in figure 4.
Figure A3: Position tests with Sc pieces in DMK13 DQ. Orientation like in figure 4.
Figure A4: Position tests with Bd3 pieces in DMK13 DQ. Orientation like in figure 4.
Figure A5: Hc results of position tests with Sc pieces in DEK 1 EQ. Three trays with test pieces on different levels of the total stack.
Figure A6: CoM results of position tests with Sc pieces in DEK 1 EQ. Three trays with test pieces on different levels of the total stack.
Figure A7: Hc/CoM for samples on 16 trays stacked vertically from the bottom of DEK1 EQ. From the same sintering as part 1 EQ test pieces.
Figure A8: Hc/CoM results of long duration ageing of Aa pieces in air.
Figure A9: Hc/CoM results of long duration ageing of Aa pieces in the argon cabinet.
Figure A10: Hc/CoM results of long duration ageing of Ab pieces in air.
Figure A12: Hc/CoM results of long duration ageing of Ac pieces in air.
Figure A13: Hc/CoM results of long duration ageing of Ac pieces in the argon cabinet.
Figure A14: Hc/CoM results of long duration ageing of Ad pieces in air.
Figure A16: Hc/CoM results of long duration ageing of Ba pieces in air.
Figure A17: Hc/CoM results of long duration ageing of Ba pieces in the argon cabinet.
Figure A18: Hc/CoM results of long duration ageing of Bb pieces in air.
Figure A20: Hc/CoM results of long duration ageing of Bc pieces in air.
Figure A21: Hc/CoM results of long duration ageing of Bc pieces in the argon cabinet.
Figure A22: Hc/CoM results of long duration ageing of Bd pieces in air.
Figure A24: Hc results for humidity tests on Bd pieces.
Figure A25: CoM results for humidity tests on Bd pieces.
Figure A26: CoM results for early stage ageing of Aa2 pieces. DMK furnace used.
Figure A27: CoM results for early stage ageing of Aa3 pieces. DMK furnace used.
Figure A28: CoM results for early stage ageing of Bd2 pieces. DMK furnace used.
Figure A29: CoM results for early stage ageing of Bd3 pieces. DMK furnace used.
Different time scale on x-axis because of longer testing.
Figure A30: Hc results for early stage ageing of Aa2 pieces. DMK furnace used, so a lot of noise purely based on orientation of the tray.
Figure A31: Hc results for early stage ageing of Aa3 pieces. DMK furnace used, so a
Figure A32: Hc results for early stage ageing of Bd2 pieces. DMK furnace used, so a lot of noise purely based on orientation of the tray.
Figure A33: Hc results for early stage ageing of Bd3 pieces. DMK furnace used, so a lot of noise purely based on orientation of the tray.
